Open by Design: Campus Fryslân's new MSc Voice Technology
|Date:||08 September 2021|
In September 2021, Campus Fryslân launched a new MSc programme in Voice Technology which is unique in Europe. The programme is not only cutting-edge and interdisciplinary in its approach to speech technologies, but also fully committed to Open Science principles. Dr Matt Coler, Associate Professor at Campus Fryslân and Director of the MSc, tells us why and how Open Science is being integrated in the programme.
The new MSc Voice Technology is committed to open science in its design. How are open science practices integrated in the curriculum?
The open access and open science backbone of the programme is evident in every facet of the curriculum. For example, nearly all assigned readings are from open access publications. The computational aspects of the programme revolve around open source code. And analyses are performed on software that doesn’t require vendor lock-in. Even the final thesis projects will be available on public repositories, freely accessible to anyone (including the next generation of students) who want to test them out and develop them further.
Why did you decide to integrate open science practices in the programme?
The teachers in the programme are committed to the idea that publicly funded education should be aligned with open science. After all, knowledge should be free -- not just for the good of science and innovation, but for the good of humanity. We believe that universities have a particular obligation in this regard. We also think that our students should learn the intrinsic value of sharing knowledge, fostering collaboration, and working together to improve the world as a good in itself!
Is openness common in the voice technology sector?
Not yet. For one thing, many voice technology innovations arise in industry, which must maintain its intellectual property to maximize profit. Moreover, data used in voice technologies involve human speech, which is, in some sense, uniquely personal and intimate. For these reasons, among others, voice technology is not very open -- and even, in some cases, adverse to the values of open science.
However, university research is not constrained by profit motives, and the human speech our students gather will be used for collaborative, experimental projects that take into account ethical considerations.
The MSc is developed in close collaboration with tech companies. How was the integration of open science practices in the course received by industry partners?
First and foremost, not all tech companies operate in opposition to open science practices. Consider the case of the Mozilla Common Voice initiative and the start-up Coqui; both are private companies which invoke a commitment to open science, collaborative practices (while still generating a profit, of course!). Alignment with these partners comes easily.
Of course we have other partners which come from more traditional Silicon Valley backgrounds, including colleagues from Google and Apple. An open science approach also works well with these companies as we do not commercialize research output and therefore don’t have any remote capacity to even slightly threaten their own research and development (R&D). That said, collaboration with these partners is hindered by the fact that they are not allowed to provide scientific support on student work for intellectual property reasons. Nonetheless, they ensure that the direction of the programme is aligned with industry expectations so our graduates can compete on the job market.
I also want to emphasize that, from our perspective, open access and open science are not incompatible with entrepreneurship. If some of the MSc students wish to commercialize their own research outcomes after earning their diploma, we have partnered with Founded In Fryslan to launch a start-up accelerator and incubation program where students can build on their open science outcomes in their own time and deliver a market-ready product.
How is this open approach perceived by the programme’s teachers and prospective students?
The instructors are aligned with the open approach -- even the one who was recruited from Samsung R&D! We all share a similar vision. The best way to illustrate this shared philosophy is by asking a few of the teachers. I share their responses below.
According to Assistant Prof. Vass Verkhodanova, “Open science is not only open access articles, available data or open source code. It starts with the community that is ready to collaborate, disseminate knowledge, and focus on accessibility, replicability, and re-usability of the research output. This is the community that we aim to build here. As an example, in my courses I promote the awareness of the open science philosophy, including the use of open source tools for speech analysis and the "FAIR" data principles”
The same holds for the programming course. The instructor, Phat Do notes that “students learn that the replicability and re-usability of their code should always be prioritized. Also, a significant portion of the course is dedicated to using and building on open-source code from others and making the students’ own code publicly available for dissemination. Moreover, the course exclusively uses freely available material, and is designed so that all students can maximize their learning without having to worry about prohibitive hardware and software costs.”
The newest instructor to join the team, Assistant Prof. Shekhar Nayak, provides an overarching summary of what open access means for Voice Technology: “Speech technology research is growing at dramatic pace due to the availability of freely open datasets like Librispeech and open source tools like Kaldi and Festival. In the MSc Voice Technology program, we promote the use of open speech and language resources such as openSLR for teaching and research. Moreover, students are highly encouraged to make open access to any relevant original data/resources collected or original codes/tools emanating from their research to contribute back to the community.”
And as for students, the open access and open science component is heavily emphasized in promotional materials and events. Many prospective students ask what kind of laptop and programs they need to buy and install before class begins. When I tell them that any laptop will do and that all programs are free, many are surprised and relieved. This approach guarantees that even after they graduate they can still access the content from class, replicate their research on different computers, and even continue their own experiments autonomously.
Could this programme be a model for other programmes to incorporate open science practices?
Since it is only year-1 of the programme, maybe it’s too early to say. We would love to work with other programs to explore different ways to collaborate, share best practices, and connect on the shared goals of advancing science and improving the quality of life for everyone!